부산대학교 도서관

The electricity demand for water supply purposes is steadily increasing due to urbanization. Therefore, water distribution networks (WDNs) are becoming energy-intensive due to the wide-spread deployment of electricity-driven water pumps. Energy-efficient operation of water pumps is a significant concern for WDN operators. To this end, this paper formulates the optimal water flow (OWF) problem to optimally schedule pumps and valves with the objective of minimizing the pump power consumption while at the same time accounting for the flow-dependent pump efficiency in WDNs. The resulting OWF problem is a mixed-integer nonlinear program (MINLP). The problem includes a fractional (nonconvex) objective and nonconvex constraints due to the WDN hydraulics, and is hard to solve. A novel successive linear approximation-based approach is used to overcome the nonconvex hydraulic constraints. Furthermore, Dinkelbach’s algorithm is used to tackle the fractional pump power objective. Finally, a solver called convex optimal water flow (C-OWF) is developed, which relies on solving a sequence of mixed-integer linear programs. A case study verified by simulation software EPANET illustrates the C-OWF’s benefits in operating the pump near maximum efficiency and reducing pump power compared to conventional rule-based designs.